Message ID | 20221020172643.3458767-4-namhyung@kernel.org |
---|---|
State | New |
Headers |
Return-Path: <linux-kernel-owner@vger.kernel.org> Delivered-To: ouuuleilei@gmail.com Received: by 2002:a5d:4242:0:0:0:0:0 with SMTP id s2csp233003wrr; Thu, 20 Oct 2022 10:27:47 -0700 (PDT) X-Google-Smtp-Source: AMsMyM5Ms5oKEEwk2dewbmniJn9lTdYc9SPL1vpMV30Bt64mMT55wZeJlTsJa8atdMT2KbUEmQbF X-Received: by 2002:a17:907:3e9c:b0:78d:ce8b:8548 with SMTP id hs28-20020a1709073e9c00b0078dce8b8548mr12153698ejc.427.1666286867601; Thu, 20 Oct 2022 10:27:47 -0700 (PDT) ARC-Seal: i=1; a=rsa-sha256; t=1666286867; cv=none; d=google.com; s=arc-20160816; b=gyuA/DZvo2jK9A6fX8I8fWOy1zDY4Qx/G0dbQ1aCgmS4HpqpbHPQbbJz9I6/3rXhj2 eZmJhb6FN1MVlhi9ORtdQC+jP3sEeqTK/De8dTQoEO29CQZd0UYG4EERdguHp/lISvuR 06VCoxZIa97iCggtkfkD3kITqVpqPHd/DQhVGVSrPx+b3hNvrgpwWGQ5I1gm/N/pUIU2 wx6OIdSw7e1JUarXwrH9oRh3PYKpLkl1xFPnRZjfLPbvBDL1MaIlpB6gbzqFvawKzG2g jw1ZjWsJWHOCoFLtu+QJBW3PdYaJcq9jk0E8uZ/urc6CYkLg8j60SjJcdPbhLuYjhVmu Mtfg== ARC-Message-Signature: i=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=arc-20160816; h=list-id:precedence:content-transfer-encoding:mime-version :references:in-reply-to:message-id:date:subject:cc:to:from:sender :dkim-signature; bh=PH0g+SRm+dkCNaIciH1RdMRbmhrFmdajvY8xociCPdQ=; b=pF2mXZl+xE6c+2lybvNJytDR7gAetw3mxMyR+bihmVVbP1VB2AYWrdJl59XmpoEu/6 crJApL5ZRjN2A2xB4eCtTxaafEcJ1IC27GlV0vmut8ITAF16fctJyQ+YarlNouu4nqyu BgLPxIOIFMOx2xst6BFbcH1DLJxfGwG2qZiED9rlZy5txAYVLhOTdQzPYznNdm9bDgtB jgJVIYPBbbaffLxz59VyfVb6ajX/ikpsX6lYYvfQFwLPND+Byd9YnFIdhV0R1XzHRqzc pO9Gw/wNtRIVm8POJ+PjH5zjh3Ed74cDbltFPGbPfcPRa4Q5HRIH9dHVaZpkwgYbvRik zkCw== ARC-Authentication-Results: i=1; mx.google.com; dkim=pass header.i=@gmail.com header.s=20210112 header.b=L4AhiqLs; spf=pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) smtp.mailfrom=linux-kernel-owner@vger.kernel.org; dmarc=pass (p=NONE sp=NONE dis=NONE) header.from=kernel.org Received: from out1.vger.email (out1.vger.email. [2620:137:e000::1:20]) by mx.google.com with ESMTP id q17-20020a1709063d5100b00781df60a2c8si13830765ejf.342.2022.10.20.10.27.18; Thu, 20 Oct 2022 10:27:47 -0700 (PDT) Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20; Authentication-Results: mx.google.com; dkim=pass header.i=@gmail.com header.s=20210112 header.b=L4AhiqLs; spf=pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) smtp.mailfrom=linux-kernel-owner@vger.kernel.org; dmarc=pass (p=NONE sp=NONE dis=NONE) header.from=kernel.org Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S229933AbiJTR1B (ORCPT <rfc822;ruipengqi7@gmail.com> + 99 others); Thu, 20 Oct 2022 13:27:01 -0400 Received: from lindbergh.monkeyblade.net ([23.128.96.19]:55860 "EHLO lindbergh.monkeyblade.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S229834AbiJTR0v (ORCPT <rfc822;linux-kernel@vger.kernel.org>); Thu, 20 Oct 2022 13:26:51 -0400 Received: from mail-pl1-x634.google.com (mail-pl1-x634.google.com [IPv6:2607:f8b0:4864:20::634]) by lindbergh.monkeyblade.net (Postfix) with ESMTPS id 8F7171A1B01; Thu, 20 Oct 2022 10:26:50 -0700 (PDT) Received: by mail-pl1-x634.google.com with SMTP id k9so20474769pll.11; Thu, 20 Oct 2022 10:26:50 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=20210112; h=content-transfer-encoding:mime-version:references:in-reply-to :message-id:date:subject:cc:to:from:sender:from:to:cc:subject:date :message-id:reply-to; bh=PH0g+SRm+dkCNaIciH1RdMRbmhrFmdajvY8xociCPdQ=; b=L4AhiqLscNtbMUNvjtXn237HAUbTRwwvcT2SyM7nA64Vng9WA6lcrXaMqGt4+MITFP fqx9GLLV/YN8HfZWeVaVTavPxjd4UaJrhkdPRd97yrJY0D1n0meMhmMlP6WRuluflecO EVM1LqI1Ya/EqcECbaHAt8TRvlJ+zcpFEWxd8U7f5apMy15Xth17x6EFh2MsCUV9aXNL rQhWOmxdUm9ZPKrmLnNXioYJo/6o3ZGzjqzitL4NMpKLWXUs48/i1KaP/VMbbwKuEOjF p91D9cqAHhXQPC/lSjU+AWA7S+YAfO9KG6QLUxri5eKjFAClqliOI9rOijGtlFlRopzr cdDA== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20210112; h=content-transfer-encoding:mime-version:references:in-reply-to :message-id:date:subject:cc:to:from:sender:x-gm-message-state:from :to:cc:subject:date:message-id:reply-to; bh=PH0g+SRm+dkCNaIciH1RdMRbmhrFmdajvY8xociCPdQ=; b=tDUklj8JRf3scEm+bDrXFkPb4LMNX5vVhUN6YuNpKrd0kZ3vJSPpCrviHMguwEg3WO BbIIU4BsBmmBuqtodyd2yeNYI9arWWuhyaiZmmHltZik8TwUXczTiZDmH16+No5yk0XZ ZC2khdlmdp2GFSB/+BYUIkythUgnWqJfAR/HUjU97CvzQEfEBrgdYR9jxCqz7q8I2Np4 vR+qcftXv9yplI6TPILKmukiHZiVcZqxZlt4Q48pKg7MclN9b5VACIN4NjSEPdhOGmg+ HklXKCZOYjLcUi6VN66/4HSS4BrGBc1fDwiaZbmpKZhZkHZhfNUiftsk8D0ckbJrQI4X 92SA== X-Gm-Message-State: ACrzQf04RoraxGY4qIdzM/jp2SIs04+EvHIjyF/lT2nQR1/zaLmJr+ZL tKSJDwrETjeBwJLVvjEARRA= X-Received: by 2002:a17:902:720a:b0:181:150c:fcc4 with SMTP id ba10-20020a170902720a00b00181150cfcc4mr15153685plb.109.1666286810012; Thu, 20 Oct 2022 10:26:50 -0700 (PDT) Received: from youngsil.svl.corp.google.com ([2620:15c:2d4:203:637f:7dcc:4bbf:4981]) by smtp.gmail.com with ESMTPSA id q13-20020a170902f34d00b0017680faa1a8sm12951754ple.112.2022.10.20.10.26.49 (version=TLS1_3 cipher=TLS_AES_256_GCM_SHA384 bits=256/256); Thu, 20 Oct 2022 10:26:49 -0700 (PDT) Sender: Namhyung Kim <namhyung@gmail.com> From: Namhyung Kim <namhyung@kernel.org> To: Arnaldo Carvalho de Melo <acme@kernel.org>, Jiri Olsa <jolsa@kernel.org> Cc: Ingo Molnar <mingo@kernel.org>, Peter Zijlstra <peterz@infradead.org>, LKML <linux-kernel@vger.kernel.org>, Ian Rogers <irogers@google.com>, Adrian Hunter <adrian.hunter@intel.com>, linux-perf-users@vger.kernel.org Subject: [PATCH 3/8] perf test: Use a test program in perf record tests Date: Thu, 20 Oct 2022 10:26:38 -0700 Message-Id: <20221020172643.3458767-4-namhyung@kernel.org> X-Mailer: git-send-email 2.38.0.135.g90850a2211-goog In-Reply-To: <20221020172643.3458767-1-namhyung@kernel.org> References: <20221020172643.3458767-1-namhyung@kernel.org> MIME-Version: 1.0 Content-Transfer-Encoding: 8bit X-Spam-Status: No, score=-1.5 required=5.0 tests=BAYES_00,DKIM_SIGNED, DKIM_VALID,DKIM_VALID_EF,FREEMAIL_FORGED_FROMDOMAIN,FREEMAIL_FROM, HEADER_FROM_DIFFERENT_DOMAINS,RCVD_IN_DNSWL_NONE,SPF_HELO_NONE, SPF_PASS,URIBL_BLOCKED autolearn=no autolearn_force=no version=3.4.6 X-Spam-Checker-Version: SpamAssassin 3.4.6 (2021-04-09) on lindbergh.monkeyblade.net Precedence: bulk List-ID: <linux-kernel.vger.kernel.org> X-Mailing-List: linux-kernel@vger.kernel.org X-getmail-retrieved-from-mailbox: =?utf-8?q?INBOX?= X-GMAIL-THRID: =?utf-8?q?1747228418440069681?= X-GMAIL-MSGID: =?utf-8?q?1747228418440069681?= |
Series |
perf test: Improve perf record tests (v2)
|
|
Commit Message
Namhyung Kim
Oct. 20, 2022, 5:26 p.m. UTC
If the system has cc it could build a test program with two threads
and then use it for more detailed testing. Also it accepts an option
to run a thread forever to ensure multi-thread runs.
If cc is not found, it falls back to use the default value 'true'.
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
---
tools/perf/tests/shell/record.sh | 64 ++++++++++++++++++++++++++++++--
1 file changed, 60 insertions(+), 4 deletions(-)
Comments
On Thu, Oct 20, 2022 at 10:26 AM Namhyung Kim <namhyung@kernel.org> wrote: > > If the system has cc it could build a test program with two threads > and then use it for more detailed testing. Also it accepts an option > to run a thread forever to ensure multi-thread runs. > > If cc is not found, it falls back to use the default value 'true'. > > Signed-off-by: Namhyung Kim <namhyung@kernel.org> Acked-by: Ian Rogers <irogers@google.com> I wonder if these utilities should just be built into perf to avoid the cc dependency. Perhaps we can have a hidden option built into perf test. Thanks, Ian > --- > tools/perf/tests/shell/record.sh | 64 ++++++++++++++++++++++++++++++-- > 1 file changed, 60 insertions(+), 4 deletions(-) > > diff --git a/tools/perf/tests/shell/record.sh b/tools/perf/tests/shell/record.sh > index 464071462809..952981481239 100755 > --- a/tools/perf/tests/shell/record.sh > +++ b/tools/perf/tests/shell/record.sh > @@ -6,10 +6,17 @@ set -e > > err=0 > perfdata=$(mktemp /tmp/__perf_test.perf.data.XXXXX) > +testprog=$(mktemp /tmp/__perf_test.prog.XXXXXX) > +testsym="test_loop" > > cleanup() { > rm -f "${perfdata}" > rm -f "${perfdata}".old > + > + if [ "${testprog}" != "true" ]; then > + rm -f "${testprog}" > + fi > + > trap - EXIT TERM INT > } > > @@ -19,9 +26,56 @@ trap_cleanup() { > } > trap trap_cleanup EXIT TERM INT > > +build_test_program() { > + if ! [ -x "$(command -v cc)" ]; then > + # No CC found. Fall back to 'true' > + testprog=true > + testsym=true > + return > + fi > + > + echo "Build a test program" > + cat <<EOF | cc -o ${testprog} -xc - -pthread > +#include <stdio.h> > +#include <stdlib.h> > +#include <pthread.h> > + > +void test_loop(void) { > + volatile int count = 1000000; > + > + while (count--) > + continue; > +} > + > +void *thfunc(void *arg) { > + int forever = *(int *)arg; > + > + do { > + test_loop(); > + } while (forever); > + > + return NULL; > +} > + > +int main(int argc, char *argv[]) { > + pthread_t th; > + int forever = 0; > + > + if (argc > 1) > + forever = atoi(argv[1]); > + > + pthread_create(&th, NULL, thfunc, &forever); > + test_loop(); > + pthread_join(th, NULL); > + > + return 0; > +} > +EOF > +} > + > test_per_thread() { > echo "Basic --per-thread mode test" > - if ! perf record -o /dev/null --quiet true 2> /dev/null > + if ! perf record -o /dev/null --quiet ${testprog} 2> /dev/null > then > echo "Per-thread record [Skipped event not supported]" > if [ $err -ne 1 ] > @@ -30,13 +84,13 @@ test_per_thread() { > fi > return > fi > - if ! perf record --per-thread -o "${perfdata}" true 2> /dev/null > + if ! perf record --per-thread -o "${perfdata}" ${testprog} 2> /dev/null > then > echo "Per-thread record [Failed record]" > err=1 > return > fi > - if ! perf report -i "${perfdata}" -q | grep -q true > + if ! perf report -i "${perfdata}" -q | grep -q "${testsym}" > then > echo "Per-thread record [Failed missing output]" > err=1 > @@ -62,7 +116,7 @@ test_register_capture() { > return > fi > if ! perf record -o - --intr-regs=di,r8,dx,cx -e br_inst_retired.near_call:p \ > - -c 1000 --per-thread true 2> /dev/null \ > + -c 1000 --per-thread ${testprog} 2> /dev/null \ > | perf script -F ip,sym,iregs -i - 2> /dev/null \ > | grep -q "DI:" > then > @@ -73,6 +127,8 @@ test_register_capture() { > echo "Register capture test [Success]" > } > > +build_test_program > + > test_per_thread > test_register_capture > > -- > 2.38.0.135.g90850a2211-goog >
On 20/10/22 20:26, Namhyung Kim wrote: > If the system has cc it could build a test program with two threads > and then use it for more detailed testing. Also it accepts an option > to run a thread forever to ensure multi-thread runs. > > If cc is not found, it falls back to use the default value 'true'. Reviewed-by: Adrian Hunter <adrian.hunter@intel.com> > > Signed-off-by: Namhyung Kim <namhyung@kernel.org> > --- > tools/perf/tests/shell/record.sh | 64 ++++++++++++++++++++++++++++++-- > 1 file changed, 60 insertions(+), 4 deletions(-) > > diff --git a/tools/perf/tests/shell/record.sh b/tools/perf/tests/shell/record.sh > index 464071462809..952981481239 100755 > --- a/tools/perf/tests/shell/record.sh > +++ b/tools/perf/tests/shell/record.sh > @@ -6,10 +6,17 @@ set -e > > err=0 > perfdata=$(mktemp /tmp/__perf_test.perf.data.XXXXX) > +testprog=$(mktemp /tmp/__perf_test.prog.XXXXXX) > +testsym="test_loop" > > cleanup() { > rm -f "${perfdata}" > rm -f "${perfdata}".old > + > + if [ "${testprog}" != "true" ]; then > + rm -f "${testprog}" > + fi > + > trap - EXIT TERM INT > } > > @@ -19,9 +26,56 @@ trap_cleanup() { > } > trap trap_cleanup EXIT TERM INT > > +build_test_program() { > + if ! [ -x "$(command -v cc)" ]; then > + # No CC found. Fall back to 'true' > + testprog=true > + testsym=true > + return > + fi > + > + echo "Build a test program" > + cat <<EOF | cc -o ${testprog} -xc - -pthread > +#include <stdio.h> > +#include <stdlib.h> > +#include <pthread.h> > + > +void test_loop(void) { > + volatile int count = 1000000; > + > + while (count--) > + continue; > +} > + > +void *thfunc(void *arg) { > + int forever = *(int *)arg; > + > + do { > + test_loop(); > + } while (forever); > + > + return NULL; > +} > + > +int main(int argc, char *argv[]) { > + pthread_t th; > + int forever = 0; > + > + if (argc > 1) > + forever = atoi(argv[1]); > + > + pthread_create(&th, NULL, thfunc, &forever); > + test_loop(); > + pthread_join(th, NULL); > + > + return 0; > +} > +EOF > +} > + > test_per_thread() { > echo "Basic --per-thread mode test" > - if ! perf record -o /dev/null --quiet true 2> /dev/null > + if ! perf record -o /dev/null --quiet ${testprog} 2> /dev/null > then > echo "Per-thread record [Skipped event not supported]" > if [ $err -ne 1 ] > @@ -30,13 +84,13 @@ test_per_thread() { > fi > return > fi > - if ! perf record --per-thread -o "${perfdata}" true 2> /dev/null > + if ! perf record --per-thread -o "${perfdata}" ${testprog} 2> /dev/null > then > echo "Per-thread record [Failed record]" > err=1 > return > fi > - if ! perf report -i "${perfdata}" -q | grep -q true > + if ! perf report -i "${perfdata}" -q | grep -q "${testsym}" > then > echo "Per-thread record [Failed missing output]" > err=1 > @@ -62,7 +116,7 @@ test_register_capture() { > return > fi > if ! perf record -o - --intr-regs=di,r8,dx,cx -e br_inst_retired.near_call:p \ > - -c 1000 --per-thread true 2> /dev/null \ > + -c 1000 --per-thread ${testprog} 2> /dev/null \ > | perf script -F ip,sym,iregs -i - 2> /dev/null \ > | grep -q "DI:" > then > @@ -73,6 +127,8 @@ test_register_capture() { > echo "Register capture test [Success]" > } > > +build_test_program > + > test_per_thread > test_register_capture >
Em Thu, Oct 20, 2022 at 04:52:14PM -0700, Ian Rogers escreveu: > On Thu, Oct 20, 2022 at 10:26 AM Namhyung Kim <namhyung@kernel.org> wrote: > > If the system has cc it could build a test program with two threads > > and then use it for more detailed testing. Also it accepts an option > > to run a thread forever to ensure multi-thread runs. > > > > If cc is not found, it falls back to use the default value 'true'. > > > > Signed-off-by: Namhyung Kim <namhyung@kernel.org> > > Acked-by: Ian Rogers <irogers@google.com> > > I wonder if these utilities should just be built into perf to avoid > the cc dependency. Perhaps we can have a hidden option built into perf > test. Agreed, not depending on a compiler makes 'perf test' usable in more systems, particularly production ones where we may want to check if perf is passing all tests applicable to that system. - Arnaldo > Thanks, > Ian > > > --- > > tools/perf/tests/shell/record.sh | 64 ++++++++++++++++++++++++++++++-- > > 1 file changed, 60 insertions(+), 4 deletions(-) > > > > diff --git a/tools/perf/tests/shell/record.sh b/tools/perf/tests/shell/record.sh > > index 464071462809..952981481239 100755 > > --- a/tools/perf/tests/shell/record.sh > > +++ b/tools/perf/tests/shell/record.sh > > @@ -6,10 +6,17 @@ set -e > > > > err=0 > > perfdata=$(mktemp /tmp/__perf_test.perf.data.XXXXX) > > +testprog=$(mktemp /tmp/__perf_test.prog.XXXXXX) > > +testsym="test_loop" > > > > cleanup() { > > rm -f "${perfdata}" > > rm -f "${perfdata}".old > > + > > + if [ "${testprog}" != "true" ]; then > > + rm -f "${testprog}" > > + fi > > + > > trap - EXIT TERM INT > > } > > > > @@ -19,9 +26,56 @@ trap_cleanup() { > > } > > trap trap_cleanup EXIT TERM INT > > > > +build_test_program() { > > + if ! [ -x "$(command -v cc)" ]; then > > + # No CC found. Fall back to 'true' > > + testprog=true > > + testsym=true > > + return > > + fi > > + > > + echo "Build a test program" > > + cat <<EOF | cc -o ${testprog} -xc - -pthread > > +#include <stdio.h> > > +#include <stdlib.h> > > +#include <pthread.h> > > + > > +void test_loop(void) { > > + volatile int count = 1000000; > > + > > + while (count--) > > + continue; > > +} > > + > > +void *thfunc(void *arg) { > > + int forever = *(int *)arg; > > + > > + do { > > + test_loop(); > > + } while (forever); > > + > > + return NULL; > > +} > > + > > +int main(int argc, char *argv[]) { > > + pthread_t th; > > + int forever = 0; > > + > > + if (argc > 1) > > + forever = atoi(argv[1]); > > + > > + pthread_create(&th, NULL, thfunc, &forever); > > + test_loop(); > > + pthread_join(th, NULL); > > + > > + return 0; > > +} > > +EOF > > +} > > + > > test_per_thread() { > > echo "Basic --per-thread mode test" > > - if ! perf record -o /dev/null --quiet true 2> /dev/null > > + if ! perf record -o /dev/null --quiet ${testprog} 2> /dev/null > > then > > echo "Per-thread record [Skipped event not supported]" > > if [ $err -ne 1 ] > > @@ -30,13 +84,13 @@ test_per_thread() { > > fi > > return > > fi > > - if ! perf record --per-thread -o "${perfdata}" true 2> /dev/null > > + if ! perf record --per-thread -o "${perfdata}" ${testprog} 2> /dev/null > > then > > echo "Per-thread record [Failed record]" > > err=1 > > return > > fi > > - if ! perf report -i "${perfdata}" -q | grep -q true > > + if ! perf report -i "${perfdata}" -q | grep -q "${testsym}" > > then > > echo "Per-thread record [Failed missing output]" > > err=1 > > @@ -62,7 +116,7 @@ test_register_capture() { > > return > > fi > > if ! perf record -o - --intr-regs=di,r8,dx,cx -e br_inst_retired.near_call:p \ > > - -c 1000 --per-thread true 2> /dev/null \ > > + -c 1000 --per-thread ${testprog} 2> /dev/null \ > > | perf script -F ip,sym,iregs -i - 2> /dev/null \ > > | grep -q "DI:" > > then > > @@ -73,6 +127,8 @@ test_register_capture() { > > echo "Register capture test [Success]" > > } > > > > +build_test_program > > + > > test_per_thread > > test_register_capture > > > > -- > > 2.38.0.135.g90850a2211-goog > >
On 24/10/22 14:33, Arnaldo Carvalho de Melo wrote: > Em Thu, Oct 20, 2022 at 04:52:14PM -0700, Ian Rogers escreveu: >> On Thu, Oct 20, 2022 at 10:26 AM Namhyung Kim <namhyung@kernel.org> wrote: > >>> If the system has cc it could build a test program with two threads >>> and then use it for more detailed testing. Also it accepts an option >>> to run a thread forever to ensure multi-thread runs. >>> >>> If cc is not found, it falls back to use the default value 'true'. >>> >>> Signed-off-by: Namhyung Kim <namhyung@kernel.org> >> >> Acked-by: Ian Rogers <irogers@google.com> >> >> I wonder if these utilities should just be built into perf to avoid >> the cc dependency. Perhaps we can have a hidden option built into perf >> test. > > Agreed, not depending on a compiler makes 'perf test' usable in more > systems, particularly production ones where we may want to check if perf > is passing all tests applicable to that system. I haven't seen anyone package anything except the perf executable, so I presume the only people running these tests install the source, and so need a compiler anyway.
Em Mon, Oct 24, 2022 at 05:00:14PM +0300, Adrian Hunter escreveu: > On 24/10/22 14:33, Arnaldo Carvalho de Melo wrote: > > Em Thu, Oct 20, 2022 at 04:52:14PM -0700, Ian Rogers escreveu: > >> I wonder if these utilities should just be built into perf to avoid > >> the cc dependency. Perhaps we can have a hidden option built into > >> perf test. > > Agreed, not depending on a compiler makes 'perf test' usable in more > > systems, particularly production ones where we may want to check if perf > > is passing all tests applicable to that system. > I haven't seen anyone package anything except the perf executable, so > I presume the only people running these tests install the source, and > so need a compiler anyway. Humm? [root@quaco ~]# head -3 /etc/os-release NAME="Fedora Linux" VERSION="36 (Workstation Edition)" ID=fedora [root@quaco ~]#G [root@quaco ~]# rpm -qi perf Name : perf Version : 5.19.4 Release : 200.fc36 Architecture: x86_64 Install Date: Mon 24 Oct 2022 12:57:34 PM -03 Group : Unspecified Size : 12663136 License : GPLv2 Signature : RSA/SHA256, Thu 25 Aug 2022 07:16:04 PM -03, Key ID 999f7cbf38ab71f4 Source RPM : kernel-tools-5.19.4-200.fc36.src.rpm Build Date : Thu 25 Aug 2022 06:30:42 PM -03 Build Host : bkernel02.iad2.fedoraproject.org Packager : Fedora Project Vendor : Fedora Project URL : http://www.kernel.org/ Bug URL : https://bugz.fedoraproject.org/kernel-tools Summary : Performance monitoring for the Linux kernel Description : This package contains the perf tool, which enables performance monitoring of the Linux kernel. [root@quaco ~]# [root@quaco ~]# rpm -ql perf /etc/bash_completion.d/perf /usr/bin/perf /usr/lib/.build-id /usr/lib/.build-id/0c /usr/lib/.build-id/0c/54d587cab1b533b9ab777717ddd256ff84f241 /usr/lib/.build-id/15 /usr/lib/.build-id/15/6552556be0d4ca8cffb8e002d8a67add55aed5 /usr/lib/.build-id/15/c4dfaa3861f530ef60def5506385d4b0b8924f /usr/lib/.build-id/1f /usr/lib/.build-id/1f/799c084e326516a24a647cfd5f484cd054e0a2 /usr/lib/.build-id/23 /usr/lib/.build-id/23/311fe148dd3f3d487ca5ace54ff2faa72ea8da /usr/lib/.build-id/37 /usr/lib/.build-id/37/c9ee70f321fae7fc1fc00fca91514c63f9f052 /usr/lib/.build-id/54 /usr/lib/.build-id/54/43b6bc332ad188b907fdde69ac02c9a69d158f /usr/lib/.build-id/7e /usr/lib/.build-id/7e/55e7a8e4df03137ec23b4135e205bc8eb77e05 /usr/lib/.build-id/84 /usr/lib/.build-id/84/3c3863a123d129c98fa7cb99a58d620c9e5edc /usr/lib/.build-id/86 /usr/lib/.build-id/86/d063d2f9833a447d96504e5f9e472048d12c49 /usr/lib/.build-id/98 /usr/lib/.build-id/98/a28589fde6ce95d9f3d305f7c9853333a9415b /usr/lib/.build-id/9e /usr/lib/.build-id/9e/567f55ca9c2b4867eb95d4182da07d4843976d /usr/lib/.build-id/b6 /usr/lib/.build-id/b6/f8c9a6d6e1990f8577a9fbac875f812a316a20 /usr/lib/.build-id/c0 /usr/lib/.build-id/c0/8b255f02ce8f3e90df2b6156d2d38b1221995d /usr/lib/.build-id/d5/e964c3413f52492e6355852d25b2ab23b03e38 /usr/lib/.build-id/e8 /usr/lib/.build-id/e8/2de614bf77f501a4511cb70151b98392669c77 /usr/lib64/libperf-jvmti.so /usr/libexec/perf-core /usr/libexec/perf-core/dlfilters /usr/libexec/perf-core/dlfilters/dlfilter-show-cycles.so /usr/libexec/perf-core/dlfilters/dlfilter-test-api-v0.so /usr/libexec/perf-core/perf-archive /usr/libexec/perf-core/perf-iostat /usr/libexec/perf-core/scripts /usr/libexec/perf-core/scripts/perl /usr/libexec/perf-core/scripts/perl/Perf-Trace-Util /usr/libexec/perf-core/scripts/perl/Perf-Trace-Util/lib /usr/libexec/perf-core/scripts/perl/Perf-Trace-Util/lib/Perf /usr/libexec/perf-core/scripts/perl/Perf-Trace-Util/lib/Perf/Trace /usr/libexec/perf-core/scripts/perl/Perf-Trace-Util/lib/Perf/Trace/Context.pm /usr/libexec/perf-core/scripts/perl/Perf-Trace-Util/lib/Perf/Trace/Core.pm /usr/libexec/perf-core/scripts/perl/Perf-Trace-Util/lib/Perf/Trace/Util.pm /usr/libexec/perf-core/scripts/perl/bin /usr/libexec/perf-core/scripts/perl/bin/check-perf-trace-record /usr/libexec/perf-core/scripts/perl/bin/failed-syscalls-record /usr/libexec/perf-core/scripts/perl/bin/failed-syscalls-report /usr/libexec/perf-core/scripts/perl/bin/rw-by-file-record /usr/libexec/perf-core/scripts/perl/bin/rw-by-file-report /usr/libexec/perf-core/scripts/perl/bin/rw-by-pid-record /usr/libexec/perf-core/scripts/perl/bin/rw-by-pid-report /usr/libexec/perf-core/scripts/perl/bin/rwtop-record /usr/libexec/perf-core/scripts/perl/bin/rwtop-report /usr/libexec/perf-core/scripts/perl/bin/wakeup-latency-record /usr/libexec/perf-core/scripts/perl/bin/wakeup-latency-report /usr/libexec/perf-core/scripts/perl/check-perf-trace.pl /usr/libexec/perf-core/scripts/perl/failed-syscalls.pl /usr/libexec/perf-core/scripts/perl/rw-by-file.pl /usr/libexec/perf-core/scripts/perl/rw-by-pid.pl /usr/libexec/perf-core/scripts/perl/rwtop.pl /usr/libexec/perf-core/scripts/perl/wakeup-latency.pl /usr/libexec/perf-core/scripts/python /usr/libexec/perf-core/scripts/python/Perf-Trace-Util /usr/libexec/perf-core/scripts/python/Perf-Trace-Util/lib /usr/libexec/perf-core/scripts/python/Perf-Trace-Util/lib/Perf /usr/libexec/perf-core/scripts/python/Perf-Trace-Util/lib/Perf/Trace /usr/libexec/perf-core/scripts/python/Perf-Trace-Util/lib/Perf/Trace/Core.py /usr/libexec/perf-core/scripts/python/Perf-Trace-Util/lib/Perf/Trace/EventClass.py /usr/libexec/perf-core/scripts/python/Perf-Trace-Util/lib/Perf/Trace/SchedGui.py /usr/libexec/perf-core/scripts/python/Perf-Trace-Util/lib/Perf/Trace/Util.py /usr/libexec/perf-core/scripts/python/arm-cs-trace-disasm.py /usr/libexec/perf-core/scripts/python/bin /usr/libexec/perf-core/scripts/python/bin/compaction-times-record /usr/libexec/perf-core/scripts/python/bin/compaction-times-report /usr/libexec/perf-core/scripts/python/bin/event_analyzing_sample-record /usr/libexec/perf-core/scripts/python/bin/event_analyzing_sample-report /usr/libexec/perf-core/scripts/python/bin/export-to-postgresql-record /usr/libexec/perf-core/scripts/python/bin/export-to-postgresql-report /usr/libexec/perf-core/scripts/python/bin/export-to-sqlite-record /usr/libexec/perf-core/scripts/python/bin/export-to-sqlite-report /usr/libexec/perf-core/scripts/python/bin/failed-syscalls-by-pid-record /usr/libexec/perf-core/scripts/python/bin/failed-syscalls-by-pid-report /usr/libexec/perf-core/scripts/python/bin/flamegraph-record /usr/libexec/perf-core/scripts/python/bin/flamegraph-report /usr/libexec/perf-core/scripts/python/bin/futex-contention-record /usr/libexec/perf-core/scripts/python/bin/futex-contention-report /usr/libexec/perf-core/scripts/python/bin/intel-pt-events-record /usr/libexec/perf-core/scripts/python/bin/intel-pt-events-report /usr/libexec/perf-core/scripts/python/bin/mem-phys-addr-record /usr/libexec/perf-core/scripts/python/bin/mem-phys-addr-report /usr/libexec/perf-core/scripts/python/bin/net_dropmonitor-record /usr/libexec/perf-core/scripts/python/bin/net_dropmonitor-report /usr/libexec/perf-core/scripts/python/bin/netdev-times-record /usr/libexec/perf-core/scripts/python/bin/netdev-times-report /usr/libexec/perf-core/scripts/python/bin/powerpc-hcalls-record /usr/libexec/perf-core/scripts/python/bin/powerpc-hcalls-report /usr/libexec/perf-core/scripts/python/bin/sched-migration-record /usr/libexec/perf-core/scripts/python/bin/sched-migration-report /usr/libexec/perf-core/scripts/python/bin/sctop-record /usr/libexec/perf-core/scripts/python/bin/sctop-report /usr/libexec/perf-core/scripts/python/bin/stackcollapse-record /usr/libexec/perf-core/scripts/python/bin/stackcollapse-report /usr/libexec/perf-core/scripts/python/bin/syscall-counts-by-pid-record /usr/libexec/perf-core/scripts/python/bin/syscall-counts-by-pid-report /usr/libexec/perf-core/scripts/python/bin/syscall-counts-record /usr/libexec/perf-core/scripts/python/bin/syscall-counts-report /usr/libexec/perf-core/scripts/python/check-perf-trace.py /usr/libexec/perf-core/scripts/python/compaction-times.py /usr/libexec/perf-core/scripts/python/event_analyzing_sample.py /usr/libexec/perf-core/scripts/python/export-to-postgresql.py /usr/libexec/perf-core/scripts/python/export-to-sqlite.py /usr/libexec/perf-core/scripts/python/exported-sql-viewer.py /usr/libexec/perf-core/scripts/python/failed-syscalls-by-pid.py /usr/libexec/perf-core/scripts/python/flamegraph.py /usr/libexec/perf-core/scripts/python/futex-contention.py /usr/libexec/perf-core/scripts/python/intel-pt-events.py /usr/libexec/perf-core/scripts/python/libxed.py /usr/libexec/perf-core/scripts/python/mem-phys-addr.py /usr/libexec/perf-core/scripts/python/net_dropmonitor.py /usr/libexec/perf-core/scripts/python/netdev-times.py /usr/libexec/perf-core/scripts/python/powerpc-hcalls.py /usr/libexec/perf-core/scripts/python/sched-migration.py /usr/libexec/perf-core/scripts/python/sctop.py /usr/libexec/perf-core/scripts/python/stackcollapse.py /usr/libexec/perf-core/scripts/python/stat-cpi.py /usr/libexec/perf-core/scripts/python/syscall-counts-by-pid.py /usr/libexec/perf-core/scripts/python/syscall-counts.py /usr/libexec/perf-core/tests /usr/libexec/perf-core/tests/attr /usr/libexec/perf-core/tests/attr.py /usr/libexec/perf-core/tests/attr/README /usr/libexec/perf-core/tests/attr/base-record /usr/libexec/perf-core/tests/attr/base-record-spe /usr/libexec/perf-core/tests/attr/base-stat /usr/libexec/perf-core/tests/attr/system-wide-dummy /usr/libexec/perf-core/tests/attr/test-record-C0 /usr/libexec/perf-core/tests/attr/test-record-basic /usr/libexec/perf-core/tests/attr/test-record-branch-any /usr/libexec/perf-core/tests/attr/test-record-branch-filter-any /usr/libexec/perf-core/tests/attr/test-record-branch-filter-any_call /usr/libexec/perf-core/tests/attr/test-record-branch-filter-any_ret /usr/libexec/perf-core/tests/attr/test-record-branch-filter-hv /usr/libexec/perf-core/tests/attr/test-record-branch-filter-ind_call /usr/libexec/perf-core/tests/attr/test-record-branch-filter-k /usr/libexec/perf-core/tests/attr/test-record-branch-filter-u /usr/libexec/perf-core/tests/attr/test-record-count /usr/libexec/perf-core/tests/attr/test-record-data /usr/libexec/perf-core/tests/attr/test-record-freq /usr/libexec/perf-core/tests/attr/test-record-graph-default /usr/libexec/perf-core/tests/attr/test-record-graph-default-aarch64 /usr/libexec/perf-core/tests/attr/test-record-graph-dwarf /usr/libexec/perf-core/tests/attr/test-record-graph-fp /usr/libexec/perf-core/tests/attr/test-record-graph-fp-aarch64 /usr/libexec/perf-core/tests/attr/test-record-group /usr/libexec/perf-core/tests/attr/test-record-group-sampling /usr/libexec/perf-core/tests/attr/test-record-group1 /usr/libexec/perf-core/tests/attr/test-record-group2 /usr/libexec/perf-core/tests/attr/test-record-no-buffering /usr/libexec/perf-core/tests/attr/test-record-no-inherit /usr/libexec/perf-core/tests/attr/test-record-no-samples /usr/libexec/perf-core/tests/attr/test-record-period /usr/libexec/perf-core/tests/attr/test-record-pfm-period /usr/libexec/perf-core/tests/attr/test-record-raw /usr/libexec/perf-core/tests/attr/test-record-spe-period /usr/libexec/perf-core/tests/attr/test-record-spe-period-term /usr/libexec/perf-core/tests/attr/test-record-spe-physical-address /usr/libexec/perf-core/tests/attr/test-stat-C0 /usr/libexec/perf-core/tests/attr/test-stat-basic /usr/libexec/perf-core/tests/attr/test-stat-default /usr/libexec/perf-core/tests/attr/test-stat-detailed-1 /usr/libexec/perf-core/tests/attr/test-stat-detailed-2 /usr/libexec/perf-core/tests/attr/test-stat-detailed-3 /usr/libexec/perf-core/tests/attr/test-stat-group /usr/libexec/perf-core/tests/attr/test-stat-group1 /usr/libexec/perf-core/tests/attr/test-stat-no-inherit /usr/libexec/perf-core/tests/pe-file.exe /usr/libexec/perf-core/tests/pe-file.exe.debug /usr/libexec/perf-core/tests/shell /usr/libexec/perf-core/tests/shell/buildid.sh /usr/libexec/perf-core/tests/shell/daemon.sh /usr/libexec/perf-core/tests/shell/lib /usr/libexec/perf-core/tests/shell/lib/probe.sh /usr/libexec/perf-core/tests/shell/lib/probe_vfs_getname.sh /usr/libexec/perf-core/tests/shell/pipe_test.sh /usr/libexec/perf-core/tests/shell/probe_vfs_getname.sh /usr/libexec/perf-core/tests/shell/record+probe_libc_inet_pton.sh /usr/libexec/perf-core/tests/shell/record+script_probe_vfs_getname.sh /usr/libexec/perf-core/tests/shell/record+zstd_comp_decomp.sh /usr/libexec/perf-core/tests/shell/record.sh /usr/libexec/perf-core/tests/shell/record_offcpu.sh /usr/libexec/perf-core/tests/shell/stat+csv_output.sh /usr/libexec/perf-core/tests/shell/stat+csv_summary.sh /usr/libexec/perf-core/tests/shell/stat+shadow_stat.sh /usr/libexec/perf-core/tests/shell/stat.sh /usr/libexec/perf-core/tests/shell/stat_all_metricgroups.sh /usr/libexec/perf-core/tests/shell/stat_all_metrics.sh /usr/libexec/perf-core/tests/shell/stat_all_pmu.sh /usr/libexec/perf-core/tests/shell/stat_bpf_counters.sh /usr/libexec/perf-core/tests/shell/test_arm_callgraph_fp.sh /usr/libexec/perf-core/tests/shell/test_arm_coresight.sh /usr/libexec/perf-core/tests/shell/test_arm_spe.sh /usr/libexec/perf-core/tests/shell/test_arm_spe_fork.sh /usr/libexec/perf-core/tests/shell/test_intel_pt.sh /usr/libexec/perf-core/tests/shell/trace+probe_vfs_getname.sh /usr/share/doc/perf /usr/share/doc/perf-tip/tips.txt /usr/share/doc/perf/examples.txt /usr/share/licenses/perf /usr/share/licenses/perf/COPYING /usr/share/man/man1/perf-annotate.1.gz /usr/share/man/man1/perf-archive.1.gz /usr/share/man/man1/perf-arm-spe.1.gz /usr/share/man/man1/perf-bench.1.gz /usr/share/man/man1/perf-buildid-cache.1.gz /usr/share/man/man1/perf-buildid-list.1.gz /usr/share/man/man1/perf-c2c.1.gz /usr/share/man/man1/perf-config.1.gz /usr/share/man/man1/perf-daemon.1.gz /usr/share/man/man1/perf-data.1.gz /usr/share/man/man1/perf-diff.1.gz /usr/share/man/man1/perf-dlfilter.1.gz /usr/share/man/man1/perf-evlist.1.gz /usr/share/man/man1/perf-ftrace.1.gz /usr/share/man/man1/perf-help.1.gz /usr/share/man/man1/perf-inject.1.gz /usr/share/man/man1/perf-intel-pt.1.gz /usr/share/man/man1/perf-iostat.1.gz /usr/share/man/man1/perf-kallsyms.1.gz /usr/share/man/man1/perf-kmem.1.gz /usr/share/man/man1/perf-kvm.1.gz /usr/share/man/man1/perf-list.1.gz /usr/share/man/man1/perf-lock.1.gz /usr/share/man/man1/perf-mem.1.gz /usr/share/man/man1/perf-probe.1.gz /usr/share/man/man1/perf-record.1.gz /usr/share/man/man1/perf-report.1.gz /usr/share/man/man1/perf-sched.1.gz /usr/share/man/man1/perf-script-perl.1.gz /usr/share/man/man1/perf-script-python.1.gz /usr/share/man/man1/perf-script.1.gz /usr/share/man/man1/perf-stat.1.gz /usr/share/man/man1/perf-test.1.gz /usr/share/man/man1/perf-timechart.1.gz /usr/share/man/man1/perf-top.1.gz /usr/share/man/man1/perf-trace.1.gz /usr/share/man/man1/perf-version.1.gz /usr/share/man/man1/perf.1.gz /usr/share/perf-core /usr/share/perf-core/strace /usr/share/perf-core/strace/groups /usr/share/perf-core/strace/groups/file /usr/share/perf-core/strace/groups/string
On Mon, Oct 24, 2022 at 9:00 AM Arnaldo Carvalho de Melo <acme@kernel.org> wrote: > > Em Mon, Oct 24, 2022 at 05:00:14PM +0300, Adrian Hunter escreveu: > > On 24/10/22 14:33, Arnaldo Carvalho de Melo wrote: > > > Em Thu, Oct 20, 2022 at 04:52:14PM -0700, Ian Rogers escreveu: > > >> I wonder if these utilities should just be built into perf to avoid > > >> the cc dependency. Perhaps we can have a hidden option built into > > >> perf test. > > > > Agreed, not depending on a compiler makes 'perf test' usable in more > > > systems, particularly production ones where we may want to check if perf > > > is passing all tests applicable to that system. > > > I haven't seen anyone package anything except the perf executable, so > > I presume the only people running these tests install the source, and > > so need a compiler anyway. > > Humm? > > [root@quaco ~]# head -3 /etc/os-release > NAME="Fedora Linux" > VERSION="36 (Workstation Edition)" > ID=fedora > [root@quaco ~]#G > > [root@quaco ~]# rpm -qi perf > Name : perf > Version : 5.19.4 > Release : 200.fc36 > Architecture: x86_64 > Install Date: Mon 24 Oct 2022 12:57:34 PM -03 > Group : Unspecified > Size : 12663136 > License : GPLv2 > Signature : RSA/SHA256, Thu 25 Aug 2022 07:16:04 PM -03, Key ID 999f7cbf38ab71f4 > Source RPM : kernel-tools-5.19.4-200.fc36.src.rpm > Build Date : Thu 25 Aug 2022 06:30:42 PM -03 > Build Host : bkernel02.iad2.fedoraproject.org > Packager : Fedora Project > Vendor : Fedora Project > URL : http://www.kernel.org/ > Bug URL : https://bugz.fedoraproject.org/kernel-tools > Summary : Performance monitoring for the Linux kernel > Description : > This package contains the perf tool, which enables performance monitoring > of the Linux kernel. > [root@quaco ~]# > > [root@quaco ~]# rpm -ql perf > /etc/bash_completion.d/perf > /usr/bin/perf > /usr/lib/.build-id > /usr/lib/.build-id/0c > /usr/lib/.build-id/0c/54d587cab1b533b9ab777717ddd256ff84f241 > /usr/lib/.build-id/15 > /usr/lib/.build-id/15/6552556be0d4ca8cffb8e002d8a67add55aed5 > /usr/lib/.build-id/15/c4dfaa3861f530ef60def5506385d4b0b8924f > /usr/lib/.build-id/1f > /usr/lib/.build-id/1f/799c084e326516a24a647cfd5f484cd054e0a2 > /usr/lib/.build-id/23 > /usr/lib/.build-id/23/311fe148dd3f3d487ca5ace54ff2faa72ea8da > /usr/lib/.build-id/37 > /usr/lib/.build-id/37/c9ee70f321fae7fc1fc00fca91514c63f9f052 > /usr/lib/.build-id/54 > /usr/lib/.build-id/54/43b6bc332ad188b907fdde69ac02c9a69d158f > /usr/lib/.build-id/7e > /usr/lib/.build-id/7e/55e7a8e4df03137ec23b4135e205bc8eb77e05 > /usr/lib/.build-id/84 > /usr/lib/.build-id/84/3c3863a123d129c98fa7cb99a58d620c9e5edc > /usr/lib/.build-id/86 > /usr/lib/.build-id/86/d063d2f9833a447d96504e5f9e472048d12c49 > /usr/lib/.build-id/98 > /usr/lib/.build-id/98/a28589fde6ce95d9f3d305f7c9853333a9415b > /usr/lib/.build-id/9e > /usr/lib/.build-id/9e/567f55ca9c2b4867eb95d4182da07d4843976d > /usr/lib/.build-id/b6 > /usr/lib/.build-id/b6/f8c9a6d6e1990f8577a9fbac875f812a316a20 > /usr/lib/.build-id/c0 > /usr/lib/.build-id/c0/8b255f02ce8f3e90df2b6156d2d38b1221995d > /usr/lib/.build-id/d5/e964c3413f52492e6355852d25b2ab23b03e38 > /usr/lib/.build-id/e8 > /usr/lib/.build-id/e8/2de614bf77f501a4511cb70151b98392669c77 Interesting, it contains some build-ids for system libraries? Thanks, Namhyung
On Mon, Oct 24, 2022 at 4:33 AM Arnaldo Carvalho de Melo <acme@kernel.org> wrote: > > Em Thu, Oct 20, 2022 at 04:52:14PM -0700, Ian Rogers escreveu: > > On Thu, Oct 20, 2022 at 10:26 AM Namhyung Kim <namhyung@kernel.org> wrote: > > > > If the system has cc it could build a test program with two threads > > > and then use it for more detailed testing. Also it accepts an option > > > to run a thread forever to ensure multi-thread runs. > > > > > > If cc is not found, it falls back to use the default value 'true'. > > > > > > Signed-off-by: Namhyung Kim <namhyung@kernel.org> > > > > Acked-by: Ian Rogers <irogers@google.com> > > > > I wonder if these utilities should just be built into perf to avoid > > the cc dependency. Perhaps we can have a hidden option built into perf > > test. > > Agreed, not depending on a compiler makes 'perf test' usable in more > systems, particularly production ones where we may want to check if perf > is passing all tests applicable to that system. Good idea. I think we can add an option (like -w/--workload) to run the specified workload rather than executing the tests. Like below (assuming we have a workload called 'noploop'). Thoughts? $ perf stat -- perf test -w noploop Thanks, Namhyung
Em Mon, Oct 24, 2022 at 09:12:21PM -0700, Namhyung Kim escreveu: > On Mon, Oct 24, 2022 at 9:00 AM Arnaldo Carvalho de Melo <acme@kernel.org> wrote: > > Em Mon, Oct 24, 2022 at 05:00:14PM +0300, Adrian Hunter escreveu: > > > On 24/10/22 14:33, Arnaldo Carvalho de Melo wrote: > > > > Em Thu, Oct 20, 2022 at 04:52:14PM -0700, Ian Rogers escreveu: > > > >> I wonder if these utilities should just be built into perf to avoid > > > >> the cc dependency. Perhaps we can have a hidden option built into > > > >> perf test. > > > > Agreed, not depending on a compiler makes 'perf test' usable in more > > > > systems, particularly production ones where we may want to check if perf > > > > is passing all tests applicable to that system. > > > I haven't seen anyone package anything except the perf executable, so > > > I presume the only people running these tests install the source, and > > > so need a compiler anyway. > > Humm? > > [root@quaco ~]# head -3 /etc/os-release > > NAME="Fedora Linux" > > VERSION="36 (Workstation Edition)" > > ID=fedora > > [root@quaco ~]#G > > [root@quaco ~]# rpm -qi perf > > Name : perf > > Version : 5.19.4 > > Release : 200.fc36 > > Architecture: x86_64 > > Install Date: Mon 24 Oct 2022 12:57:34 PM -03 > > Group : Unspecified > > Size : 12663136 > > License : GPLv2 > > Signature : RSA/SHA256, Thu 25 Aug 2022 07:16:04 PM -03, Key ID 999f7cbf38ab71f4 > > Source RPM : kernel-tools-5.19.4-200.fc36.src.rpm > > Build Date : Thu 25 Aug 2022 06:30:42 PM -03 > > Build Host : bkernel02.iad2.fedoraproject.org > > Packager : Fedora Project > > Vendor : Fedora Project > > URL : http://www.kernel.org/ > > Bug URL : https://bugz.fedoraproject.org/kernel-tools > > Summary : Performance monitoring for the Linux kernel > > Description : > > This package contains the perf tool, which enables performance monitoring > > of the Linux kernel. > > [root@quaco ~]# > > > > [root@quaco ~]# rpm -ql perf > > /etc/bash_completion.d/perf > > /usr/bin/perf > > /usr/lib/.build-id > > /usr/lib/.build-id/0c > > /usr/lib/.build-id/0c/54d587cab1b533b9ab777717ddd256ff84f241 > > /usr/lib/.build-id/15 > > /usr/lib/.build-id/15/6552556be0d4ca8cffb8e002d8a67add55aed5 > > /usr/lib/.build-id/15/c4dfaa3861f530ef60def5506385d4b0b8924f > > /usr/lib/.build-id/1f > > /usr/lib/.build-id/1f/799c084e326516a24a647cfd5f484cd054e0a2 > > /usr/lib/.build-id/23 > > /usr/lib/.build-id/23/311fe148dd3f3d487ca5ace54ff2faa72ea8da > > /usr/lib/.build-id/37 > > /usr/lib/.build-id/37/c9ee70f321fae7fc1fc00fca91514c63f9f052 > > /usr/lib/.build-id/54 > > /usr/lib/.build-id/54/43b6bc332ad188b907fdde69ac02c9a69d158f > > /usr/lib/.build-id/7e > > /usr/lib/.build-id/7e/55e7a8e4df03137ec23b4135e205bc8eb77e05 > > /usr/lib/.build-id/84 > > /usr/lib/.build-id/84/3c3863a123d129c98fa7cb99a58d620c9e5edc > > /usr/lib/.build-id/86 > > /usr/lib/.build-id/86/d063d2f9833a447d96504e5f9e472048d12c49 > > /usr/lib/.build-id/98 > > /usr/lib/.build-id/98/a28589fde6ce95d9f3d305f7c9853333a9415b > > /usr/lib/.build-id/9e > > /usr/lib/.build-id/9e/567f55ca9c2b4867eb95d4182da07d4843976d > > /usr/lib/.build-id/b6 > > /usr/lib/.build-id/b6/f8c9a6d6e1990f8577a9fbac875f812a316a20 > > /usr/lib/.build-id/c0 > > /usr/lib/.build-id/c0/8b255f02ce8f3e90df2b6156d2d38b1221995d > > /usr/lib/.build-id/d5/e964c3413f52492e6355852d25b2ab23b03e38 > > /usr/lib/.build-id/e8 > > /usr/lib/.build-id/e8/2de614bf77f501a4511cb70151b98392669c77 > > Interesting, it contains some build-ids for system libraries? [acme@quaco ~]$ ls -la /usr/lib/.build-id/e8/2de614bf77f501a4511cb70151b98392669c77 lrwxrwxrwx. 1 root root 67 Aug 25 18:33 /usr/lib/.build-id/e8/2de614bf77f501a4511cb70151b98392669c77 -> ../../../../usr/libexec/perf-core/dlfilters/dlfilter-test-api-v0.so [acme@quaco ~]$ Yeah, strange: [acme@quaco ~]$ rpm -ql perf | grep build-id/../ /usr/lib/.build-id/0c/54d587cab1b533b9ab777717ddd256ff84f241 /usr/lib/.build-id/15/6552556be0d4ca8cffb8e002d8a67add55aed5 /usr/lib/.build-id/15/c4dfaa3861f530ef60def5506385d4b0b8924f /usr/lib/.build-id/1f/799c084e326516a24a647cfd5f484cd054e0a2 /usr/lib/.build-id/23/311fe148dd3f3d487ca5ace54ff2faa72ea8da /usr/lib/.build-id/37/c9ee70f321fae7fc1fc00fca91514c63f9f052 /usr/lib/.build-id/54/43b6bc332ad188b907fdde69ac02c9a69d158f /usr/lib/.build-id/7e/55e7a8e4df03137ec23b4135e205bc8eb77e05 /usr/lib/.build-id/84/3c3863a123d129c98fa7cb99a58d620c9e5edc /usr/lib/.build-id/86/d063d2f9833a447d96504e5f9e472048d12c49 /usr/lib/.build-id/98/a28589fde6ce95d9f3d305f7c9853333a9415b /usr/lib/.build-id/9e/567f55ca9c2b4867eb95d4182da07d4843976d /usr/lib/.build-id/b6/f8c9a6d6e1990f8577a9fbac875f812a316a20 /usr/lib/.build-id/c0/8b255f02ce8f3e90df2b6156d2d38b1221995d /usr/lib/.build-id/d5/e964c3413f52492e6355852d25b2ab23b03e38 /usr/lib/.build-id/e8/2de614bf77f501a4511cb70151b98392669c77 [acme@quaco ~]$ [acme@quaco ~]$ rpm -ql perf | grep build-id/../ | xargs realpath /usr/lib64/traceevent/plugins/plugin_sched_switch.so /usr/lib64/traceevent/plugins/plugin_kmem.so /usr/lib64/traceevent/plugins/plugin_tlb.so /usr/lib64/traceevent/plugins/plugin_jbd2.so /usr/lib64/traceevent/plugins/plugin_hrtimer.so /usr/lib64/traceevent/plugins/plugin_mac80211.so /usr/lib64/libperf-jvmti.so /usr/lib64/traceevent/plugins/plugin_cfg80211.so /usr/lib64/traceevent/plugins/plugin_xen.so /usr/lib64/traceevent/plugins/plugin_kvm.so /usr/libexec/perf-core/dlfilters/dlfilter-show-cycles.so /usr/lib64/traceevent/plugins/plugin_scsi.so /usr/lib64/traceevent/plugins/plugin_futex.so /usr/bin/perf /usr/lib64/traceevent/plugins/plugin_function.so /usr/libexec/perf-core/dlfilters/dlfilter-test-api-v0.so [acme@quaco ~]$ rpm -ql perf | grep build-id/../ | xargs realpath | xargs rpm -qf libtraceevent-1.5.3-2.fc36.x86_64 libtraceevent-1.5.3-2.fc36.x86_64 libtraceevent-1.5.3-2.fc36.x86_64 libtraceevent-1.5.3-2.fc36.x86_64 libtraceevent-1.5.3-2.fc36.x86_64 libtraceevent-1.5.3-2.fc36.x86_64 perf-5.19.4-200.fc36.x86_64 libtraceevent-1.5.3-2.fc36.x86_64 libtraceevent-1.5.3-2.fc36.x86_64 libtraceevent-1.5.3-2.fc36.x86_64 perf-5.19.4-200.fc36.x86_64 libtraceevent-1.5.3-2.fc36.x86_64 libtraceevent-1.5.3-2.fc36.x86_64 perf-5.19.4-200.fc36.x86_64 libtraceevent-1.5.3-2.fc36.x86_64 perf-5.19.4-200.fc36.x86_64 [acme@quaco ~]$ Humm, but then the build-ids are different... [acme@quaco ~]$ rpm -ql libtraceevent | grep build-id/../ /usr/lib/.build-id/00/59bd37dc1c345eb2f98cf2bc1163bc1fd4b63e /usr/lib/.build-id/00/b22aa8cc2b81e4cb383d8b4dc14cbeb958fda6 /usr/lib/.build-id/11/6a8d17c89795d2269d3d65906ea7c7c5bcb856 /usr/lib/.build-id/24/8bfb4534954eb71ab3e5f7887507cb8981f921 /usr/lib/.build-id/57/0768e8ea5155cf2759c4f5dca3736a73c3dfad /usr/lib/.build-id/61/2744cd9cb2d1b0c569b0c86578d11be7227f27 /usr/lib/.build-id/9d/630834441f4e382ddcecef810b100737bb56e4 /usr/lib/.build-id/b0/57189b558351ace427ae0a650c248deca4badd /usr/lib/.build-id/b1/a0003b83e307cf61d7c8a2751e5c78e6ccfc87 /usr/lib/.build-id/b4/b8a30ebf86cb849bcb0cabc4d452c1a44487cc /usr/lib/.build-id/ce/5e0647f756ec2c30b2e6df9a5a98e552871cb6 /usr/lib/.build-id/e0/a6a6207f56674ffacd7b063fd2acd4cec3f661 /usr/lib/.build-id/f4/86fb5b3072945405d0590045715d18adc7aa18 [acme@quaco ~]$ [acme@quaco ~]$ rpm -ql libtraceevent | grep build-id/../ | xargs realpath /usr/lib64/traceevent/plugins/plugin_function.so /usr/lib64/traceevent/plugins/plugin_scsi.so /usr/lib64/traceevent/plugins/plugin_futex.so /usr/lib64/traceevent/plugins/plugin_cfg80211.so /usr/lib64/traceevent/plugins/plugin_hrtimer.so /usr/lib64/traceevent/plugins/plugin_mac80211.so /usr/lib64/traceevent/plugins/plugin_sched_switch.so /usr/lib64/traceevent/plugins/plugin_jbd2.so /usr/lib64/traceevent/plugins/plugin_tlb.so /usr/lib64/traceevent/plugins/plugin_xen.so /usr/lib64/traceevent/plugins/plugin_kvm.so /usr/lib64/traceevent/plugins/plugin_kmem.so /usr/lib64/libtraceevent.so.1.5.3 [acme@quaco ~]$ rpm -ql libtraceevent | grep build-id/../ | xargs realpath | xargs rpm -qf libtraceevent-1.5.3-2.fc36.x86_64 libtraceevent-1.5.3-2.fc36.x86_64 libtraceevent-1.5.3-2.fc36.x86_64 libtraceevent-1.5.3-2.fc36.x86_64 libtraceevent-1.5.3-2.fc36.x86_64 libtraceevent-1.5.3-2.fc36.x86_64 libtraceevent-1.5.3-2.fc36.x86_64 libtraceevent-1.5.3-2.fc36.x86_64 libtraceevent-1.5.3-2.fc36.x86_64 libtraceevent-1.5.3-2.fc36.x86_64 libtraceevent-1.5.3-2.fc36.x86_64 libtraceevent-1.5.3-2.fc36.x86_64 libtraceevent-1.5.3-2.fc36.x86_64 [acme@quaco ~]$ I.e. libtraceevent has the plugins and the build-id symlinks, perf has just the symlinks: [acme@quaco ~]$ rpm -ql libtraceevent | grep plugin_.*so /usr/lib64/traceevent/plugins/plugin_cfg80211.so /usr/lib64/traceevent/plugins/plugin_function.so /usr/lib64/traceevent/plugins/plugin_futex.so /usr/lib64/traceevent/plugins/plugin_hrtimer.so /usr/lib64/traceevent/plugins/plugin_jbd2.so /usr/lib64/traceevent/plugins/plugin_kmem.so /usr/lib64/traceevent/plugins/plugin_kvm.so /usr/lib64/traceevent/plugins/plugin_mac80211.so /usr/lib64/traceevent/plugins/plugin_sched_switch.so /usr/lib64/traceevent/plugins/plugin_scsi.so /usr/lib64/traceevent/plugins/plugin_tlb.so /usr/lib64/traceevent/plugins/plugin_xen.so [acme@quaco ~]$ rpm -ql perf | grep plugin_.*so [acme@quaco ~]$ And perf is being built with the libtraceevent that is in fedora: [acme@quaco ~]$ ldd `which perf` | grep traceevent libtraceevent.so.1 => /lib64/libtraceevent.so.1 (0x00007f43496dc000) [acme@quaco ~]$ rpm -qf /lib64/libtraceevent.so.1 libtraceevent-1.5.3-2.fc36.x86_64 [acme@quaco ~]$ rpm -qf /usr/bin/perf perf-5.19.4-200.fc36.x86_64 [acme@quaco ~]$ [acme@quaco ~]$ file /usr/bin/perf /usr/bin/perf: ELF 64-bit LSB pie executable, x86-64, version 1 (SYSV), dynamically linked, interpreter /lib64/ld-linux-x86-64.so.2, BuildID[sha1]=c08b255f02ce8f3e90df2b6156d2d38b1221995d, for GNU/Linux 3.2.0, stripped [acme@quaco ~]$ So it seems its just a matter of removing those /usr/lib/.build-id/ symlinks to the libtraceevent plugins from the fedora perf rpm. Justin? - Arnaldo
On 24/10/22 19:00, Arnaldo Carvalho de Melo wrote: > Em Mon, Oct 24, 2022 at 05:00:14PM +0300, Adrian Hunter escreveu: >> On 24/10/22 14:33, Arnaldo Carvalho de Melo wrote: >>> Em Thu, Oct 20, 2022 at 04:52:14PM -0700, Ian Rogers escreveu: >>>> I wonder if these utilities should just be built into perf to avoid >>>> the cc dependency. Perhaps we can have a hidden option built into >>>> perf test. > >>> Agreed, not depending on a compiler makes 'perf test' usable in more >>> systems, particularly production ones where we may want to check if perf >>> is passing all tests applicable to that system. > >> I haven't seen anyone package anything except the perf executable, so >> I presume the only people running these tests install the source, and >> so need a compiler anyway. > > Humm? > > [root@quaco ~]# head -3 /etc/os-release > NAME="Fedora Linux" > VERSION="36 (Workstation Edition)" > ID=fedora > [root@quaco ~]#G I guess I got confused - it seems it is only Ubuntu that does that.
diff --git a/tools/perf/tests/shell/record.sh b/tools/perf/tests/shell/record.sh index 464071462809..952981481239 100755 --- a/tools/perf/tests/shell/record.sh +++ b/tools/perf/tests/shell/record.sh @@ -6,10 +6,17 @@ set -e err=0 perfdata=$(mktemp /tmp/__perf_test.perf.data.XXXXX) +testprog=$(mktemp /tmp/__perf_test.prog.XXXXXX) +testsym="test_loop" cleanup() { rm -f "${perfdata}" rm -f "${perfdata}".old + + if [ "${testprog}" != "true" ]; then + rm -f "${testprog}" + fi + trap - EXIT TERM INT } @@ -19,9 +26,56 @@ trap_cleanup() { } trap trap_cleanup EXIT TERM INT +build_test_program() { + if ! [ -x "$(command -v cc)" ]; then + # No CC found. Fall back to 'true' + testprog=true + testsym=true + return + fi + + echo "Build a test program" + cat <<EOF | cc -o ${testprog} -xc - -pthread +#include <stdio.h> +#include <stdlib.h> +#include <pthread.h> + +void test_loop(void) { + volatile int count = 1000000; + + while (count--) + continue; +} + +void *thfunc(void *arg) { + int forever = *(int *)arg; + + do { + test_loop(); + } while (forever); + + return NULL; +} + +int main(int argc, char *argv[]) { + pthread_t th; + int forever = 0; + + if (argc > 1) + forever = atoi(argv[1]); + + pthread_create(&th, NULL, thfunc, &forever); + test_loop(); + pthread_join(th, NULL); + + return 0; +} +EOF +} + test_per_thread() { echo "Basic --per-thread mode test" - if ! perf record -o /dev/null --quiet true 2> /dev/null + if ! perf record -o /dev/null --quiet ${testprog} 2> /dev/null then echo "Per-thread record [Skipped event not supported]" if [ $err -ne 1 ] @@ -30,13 +84,13 @@ test_per_thread() { fi return fi - if ! perf record --per-thread -o "${perfdata}" true 2> /dev/null + if ! perf record --per-thread -o "${perfdata}" ${testprog} 2> /dev/null then echo "Per-thread record [Failed record]" err=1 return fi - if ! perf report -i "${perfdata}" -q | grep -q true + if ! perf report -i "${perfdata}" -q | grep -q "${testsym}" then echo "Per-thread record [Failed missing output]" err=1 @@ -62,7 +116,7 @@ test_register_capture() { return fi if ! perf record -o - --intr-regs=di,r8,dx,cx -e br_inst_retired.near_call:p \ - -c 1000 --per-thread true 2> /dev/null \ + -c 1000 --per-thread ${testprog} 2> /dev/null \ | perf script -F ip,sym,iregs -i - 2> /dev/null \ | grep -q "DI:" then @@ -73,6 +127,8 @@ test_register_capture() { echo "Register capture test [Success]" } +build_test_program + test_per_thread test_register_capture